|
Post by tachyon on May 7, 2022 7:10:09 GMT
Is he a target man? I don’t really know much about him. 11 goals as a target man in this league would have been a decent haul. 11, 13, 31, 11 is his goal scoring record the past 4 seasons, and reads the same backwards. Is the 31 a blip? This was in a promotion winning League One team, the 11 in a relegated Championship team. If he is playing in a promotion challenging team in the Championship, you would hope that it would be better than 11, but probably not as high as 31. With my analysts hat on. Clarke-Harris is 27 and 300 days, which is at the top end of peak & near to the start of a striker's ageing phase. Before this season he hadn't played at Championship level since 2016/17, when he barely played for a relegated team. Biggest red flag is his penalty scoring record. It has contributed 2, 4, 7, 5 since 2018/19, He's only scored 6 non penalty goals in 32.5 full games this season or 0.18 per 90. Here's our ratings for Championship strikers from this season. Caveat, some have played very few minutes and Opta's definition of a striker can be fluid. Attachment Deleted
|
|
|
Post by tachyon on May 1, 2022 18:33:41 GMT
Surridge was a non penalty expected goal every 2 games striker at Swansea & Bournemouth in a combined 2099 minutes. Play him for just 786 and 610 mins respectively at Stoke & NF and you're going to get wider performance error bars (for better or worse) and good or bad variance in his actual scoring rate. We went 1 from 2,(getting some streaky cold finishing) while NF are currently 2 from 2, (getting some very hot, streaky finishing) although they are stuffing the ball into the six yard box more frequently than we did, despite the numerous clues at Swans & AFCB. He's an above average Championship striker/poacher. His shot maps are very good. Attachment Deleted
|
|
|
Post by tachyon on May 1, 2022 17:49:44 GMT
He’s a complete basket case but Forest fans rate him highly. It’s pretty telling that the top 6 on that list play for current top 6 sides. Could imply that goalies' stats depend very much on who is in front of them and how consistently well they play? You can isolate shot stopping performance by looking at the difficulty of the shots they face. A screamer in the top corner is harder to save than a 30 yarder, straight down the keeper's throat, hence save percentage is stats trivia, not stats analysis. Based on save difficulty, Samba has the 11th best shot stopping performance this season. You need a sample size of around 100 shots on target faced for true talent, rather than random luck to begin to come to the fore.
|
|
|
Post by tachyon on May 1, 2022 17:11:19 GMT
Based on the quality of the on target shots he's faced in the Championship (~100) he's allowed around 20% more goals than an average Championship keeper. That ranks him 147th out of 162 for keepers who have faced at least 20 shots since 2014. At 21 and 300 days he needs to start moving the dial. I'm not his biggest fan, but he seems okay, so could this below average performance be due to his age? What's he like compared to his peer group through the years? We possibly let go of Bachmann too early, maybe Bursik could be another. Keepers under the age of 22 (Bursik's age) typically make up around 16% of keepers who get at least one start, but the only play 9% of the minutes once you correct for there only being one keeper. So they do get more minutes as they move into the peak age, (which ends later than other positions). That's in Bursik's favour, but if he does improve it is from a very low baseline. Also, 40% of players whose shot stopping is as "good" or worse than Bursik's current level make very few appearances in the subsequent season. You really want to see more.
|
|
|
Post by tachyon on May 1, 2022 9:54:41 GMT
The stats show that when he plays, he’s an average Champo keeper, which given his age and the level he’s playing at, is very impressive. He’ll continue to learn and get better, provided we stick with him. Based on the quality of the on target shots he's faced in the Championship (~100) he's allowed around 20% more goals than an average Championship keeper. That ranks him 147th out of 162 for keepers who have faced at least 20 shots since 2014. At 21 and 300 days he needs to start moving the dial.
|
|
|
Post by tachyon on Apr 30, 2022 6:46:42 GMT
Can you afford a wing back who doesn't take his man on? For as much as Tymon has done this season, Smith has still got twice as many assists and the same number of goals in less games. How important is taking a man on? Personally, I think that Smith in the side with Powell and Brown works really well. Ideally, we would get a more explosive right wing back who isn’t a mess defensively, but if one isn’t available, I’d rather keep Smith. As a fully paid up member of the "underlying stats trump actual output" because the former has better sample size & the latter is prone to variance, I'd say..... TS has been the final passer before a chance is attempted on 31 occasions, but only 19 have come in open play. JT has created 47 chances, 46 from open play. So TS has an opportunity based advantage because he takes set pieces (nothing wrong with that, but quick check, they account for over half of his actual assists). TS has 2.2 expected assists from open play, JT has 6.9 xA TS has 1.3 xA from set pieces, JT has 0.08 xA Account for minutes played and JT is a much better open play creative player. JT created 0.17 xA per full game from open play, TS created 0.08 (half JT's rate). JT created 1.13 open play chances per full game, TS created 0.66 (almost half JT's rate). JT created open play chances worth 0.15 xA per chance created, TS created 0.12 xA per chance (lower quality of chances compared to JT). Stylistically, JT plays infield much more than TS, who hugs the touchline a bit too much to open up space in the attacking third. (No hatchet job intended).
|
|
|
Post by tachyon on Apr 29, 2022 20:37:40 GMT
Around 35% of our crosses are from very good initial positions and around half of those 35% find a good end point in the box. So overall we get ~18% of our crosses from the optimum start point to the optimum end point. That's better than league average. Tymon especially good. It's the difference between an expected conversion rate of ~5% for only really good choices, to 1% for an average mix of crosses to ~ 0.2% conversion rates for universally bad choices. Cheers Interesting. If O'Neill uses this kind of data, it doesn't bode well for Tommy Smith's future. Tymon's a better chance creator and a smarter crosser, but the huge difference is that Tymon also attempts over three times more take on's per game. He' multi dimesional in the attacking third.
|
|
|
Post by tachyon on Apr 29, 2022 20:06:05 GMT
You can improve conversion rates by crossing from smart positions. Get as deep as the 18 yard line or deeper & cross from infield not the touchline, group your crosses between the pen spot & the six yard box, go far, rather than near post. Improves the second phase pickings as well. Do you know to what extent we have done that this season? Around 35% of our crosses are from very good initial positions and around half of those 35% find a good end point in the box. So overall we get ~18% of our crosses from the optimum start point to the optimum end point. That's better than league average. Tymon especially good. It's the difference between an expected conversion rate of ~5% for only really good choices, to 1% for an average mix of crosses to ~ 0.2% conversion rates for universally bad choices.
|
|
|
Post by tachyon on Apr 29, 2022 17:54:38 GMT
Thanks very much. So as a team we have done a decent job of getting on the end of crosses. Maybe those low crosses that nobody seemed to attack have just stuck in the memory more, even if they're not reflective of our overall performance. You can improve conversion rates by crossing from smart positions. Get as deep as the 18 yard line or deeper & cross from infield not the touchline, group your crosses between the pen spot & the six yard box, go far, rather than near post. Improves the second phase pickings as well.
|
|
|
Post by tachyon on Apr 29, 2022 16:56:44 GMT
A few times we've seen Tymon put crosses into great areas, but Maja has either looked hesitant, or just not seemed to anticipate the chance quickly enough. If I was to pick one simple way we could score more goals, it would be to sign a "meat and potatoes" striker who can get on the end of those crosses. Do the stats back up what my eyes are telling me? And do the stats show whether Maja can be that man? [/quote] You score an average of one goal directly as a result of a cross every 95 ish deliveries. If you look beyond the cross and include goals scored upto the defending side taking control of the ball you reduce those 95 crosses down to just under 50. As a team, Stoke this season have been around that historical average at converting crosses. We've scored nine goals from 870 deliveries, with seven different players providing the assist. Vrancic, Wilmot, Sawyers, Baker and Clucas & two each from Smith and Tymon. We make 5 accurate crosses per full game (our player gets the first touch). We're the 8th most frequent crossing team and 7th most efficient at getting first contact. So it's a big part of our attacking strategy. Re Maja. Fair to say heading isn't his strength (he hasn't scored a Champ, Ligue 1 or PL header. He may have done for Sunderland in 2018/19 in League one, but that was a season that data largely forgot). But his shot map does look like goal poacher, for want of a better term and Powell and Brown are sensationally good at scoring headed goals, which may compensate. Brown also excelled at providing scoring chances to feet at Barnsley (which is much more Maja's thing). Maja's play outside the box is fine and he's only played 991 mins for Stoke compared to 3505 mins at Championship level or higher for others when his attacking xG stats were fine. So you could perhaps cut him some adjustment slack. Price point will decide, I suppose.
|
|
|
Post by tachyon on Apr 29, 2022 8:56:11 GMT
Maja's shot profile in a variety of leagues (PL Ligue1 Champ) Overall 0.29 non penalty xG per full game (a NP expected goal every 3.5 games). 0.16 xG per goal attempt (so, high quality chances, likes the six yard box). Average of 1.8 attempts per game. Not done so well at Stoke (service probably an issue). Here he's got on the end of an expected goal every five full games, at roughly the same quality of chances, but fewer attempts per full game (just 1.2). Stylistically, he's poor in the air. Doesn't get involved in many aerial duels and only wins a quarter of the ones he does. (average success rate for his position is just over 40%) Passing wise, he gets more involved in passes in his own half than the average for forwards and is above average for completion, not a liability in his own half. Likes a backwards pass (14% more frequently than average for his position) So brings in midfield runners from deep and does it at above average efficiency. A smaller percentage of his passes are forward ones compared to his position and that's a career trait. Doesn't carry the ball much, but is more difficult to shift off it than average when he does. Not yet in his prime, still a young talent at 23 & 100 or so days.
|
|
|
Post by tachyon on Apr 28, 2022 11:53:48 GMT
To you maybe but to anyone seeking a reliable assessment of a player's actual abilities - stick to the data. It's what anyone who actually has responsibility for making decisions in professional football is going to do anyway. And for good reason. If these are the results, the "good reason" doesn't seem to be that good. For those still remotely interested, here's what's happening. 1) Bayern has found some stats based trivia relating to Ben Wilmott, amongst others. 2) This trivia perhaps identifies BW as a progressive passer. ( BO won't reveal his source). 3) BW does actually pass the ball forward 15% more frequently than the average Championship defender who also play in his positions. 4) But BO has then conflated this piece of trivia with BW being a good progressor of the ball. 5) An even slightly more detailed analysis shows that BW's completion rate for progressive passes is 18% BELOW that for other centre backs.. 6) So without doing a full blown passing model for BW, we can fairly confidently assume that he is a prolific, but inefficient progressor of the ball. 7) Bookmark this post, because whenever data analysis is mentioned BO will pile in with a "remember when data said BW was a good ball progessor". It actual doesn't. (Seen this anti data tactic 100's of times).
|
|
|
Post by tachyon on Apr 27, 2022 10:22:14 GMT
away from Baker, Interested to know how you compare the relative quality of leagues? Or is that the secret sauce? You get a fair bit of movenent of players between leagues and you get lots of inter league games in the European competitions. You use those to estimate exhange rates for goals etc.
|
|
|
Post by tachyon on Apr 27, 2022 9:18:25 GMT
But it’s illogical to assume that he’s going to carry on this run of goal scoring next season. He’d be on track to notch about 19 or 20 next season with his current rate. As a number 8, that just seems very unlikely. It's a legitimate concern. Firstly you always deduct penalty goals to see what kind of scoring contribution you've got. Pens go in at 80%. I know it's Stoke, but nevertheless. Here's his shot map in the Championship, with and without Stoke. Long rangers, not much in the way of big chances. He's running very, very hot for Stoke, 5 non pen goals from 2.3 xG. Project his actual career Championship non penalty scoring record over a FULL 46 game season and you get around 7 goals. Do it for his Championship non penalty xG and you get just 5 or 6 goals for an uninterrupted campaign. He projects to 10 Non pen goals based on his two Eredivisie seasons, (which is Championship comparable), but it also has 16% more goals scored, so that bumps him down to around 8. Not a Jacob Brown in waiting just yet.
|
|
|
Post by tachyon on Apr 26, 2022 12:02:23 GMT
When Brown was being castigated last season,didn't Tachyon say that the data showed that he was a good investment. Doesn't the data suggest that Brown is overachieving at the moment and may not replicate or improve on his form next year? Some data wins I've posted. Stoke will be out of the bottom three by Christmas when we were effectively rock bottom after 10. We made it on Boxing Day iirc. Who are the best free agent keepers when we were short in that position 1n 2020/21. Lonergan and Maenpaa were rated 1st & 2nd. We signed them both. Clucas' goals fall off after his mega over-performing season (Fish in a barrel). And yes JB is an excellent raw talent, who you could spot a mile off at Barnsley using data. Particularly enticing because he'd only scored 3 from 8 xG (all open play, no pens, free kicks) & the 60 odd chances he'd made for teammates were huge because he presented every single one to the players feet, rather than their head. His experience to age was amazingly mature for a young player and availability is a often ignored asset and he's been universally available to play. Yes he is slightly over achieving at the moment 13 Championship goals from 10xG (Hot, but not Tyrese worryingly hot). I get the "trust my eyes" stuff, although it appears not to apply when the ball ends up in the net 20 times in just under 2 seasons. semi rant over :-)
|
|
|
Post by tachyon on Apr 26, 2022 9:30:14 GMT
My eyes and the eyes of thousands of supporters are worth more. Neither are progressive passers of a ball. [/quote][/b] As you want to bash data with the "BW as a progressive passer" stick, can you point out where exactly you've sourced this "data" claim.
|
|
|
Post by tachyon on Apr 26, 2022 8:20:28 GMT
This is how you use data to find players who progresses the ball.
1) You value every blade of grass. (If you have possession in your own box, the chances of your team ultimately scoring from there is small, if you have possession in the opposition box, your chances are bigger).
2) If you move the ball forward from point A to point B, you have improved the chances of scoring and that can be measure as the difference in the value of the start point of your action and the successful end point. (Distance advanced and value added isn't connected. Moving 5 yards forward from a defensive position is worth much less than moving 5 yards forward from your attacking third).
3) A player can progess the ball by passing it or carrying it.
4) The likelihood of a player completing a pass can easily be modelled, based on where the pass starts and where it ends, as well as other variable such as phase of play & if it is a headed or kicked pass.
5) For any player you can therefore see if they are completing more or fewer passes than expected based on the difficulty of those pass attempts (the probability bit of the model) and you can calculate how much they improve the chances of their team scoring per pass attempt and per forward carry. (the value bit of the model). You can also see how much they hurt their team by incompleted passes or being dispossessed.
If the "evidence" you are quoting doesn't include these aspects, it's just trivia in, trivia out. So when using data to make a point, quote your methodology or primary source, otherwise it's worthless.
|
|
|
Post by tachyon on Apr 25, 2022 17:31:39 GMT
The data is collected by a real person, in real time, assisted by automated computer software.
It's checked again post game.
A Premier League title winning team independently rechecked a whole seasons worth of such data, which they had purchased from a major content provider.
They were happy with the quality control.
It is very unlikely that the data is compromised, particularly the dataset that includes individual passing events.
But what is much more likely is that Bayern has been clickbaited by a superficial scatterplot or Venn Diagram.
The source would be handy :-)
|
|
|
Post by tachyon on Apr 25, 2022 8:16:50 GMT
Fascinating insight into how the game changes because of data led insights. Particularly interesting comments re Tymon. Crossing from area A to area B are the killer zones for a cross. Conversion rates are around 6 to 7%. Tymon gets in and around zone A frequently. If he goes aerial we've got two of the division's best headers of a ball in Powell & Brown. Also both very intelligent at finding space in congested areas. Attachment Deleted
|
|
|
Post by tachyon on Apr 25, 2022 7:43:44 GMT
Can't imasgine Real Madrid pay much attention to that curve. Their three midfielders are 100 years old and it's the best midfield in the world. Yep, the way RM have maintained the same level of goals scored and allowed in Europe's premier competition compared to when Modric, Kroos and Casemiro were skipping about in their 20's instead of going into a persistent decline as their 100 year old midfield grew old is hugely impressive. If only it were true :-) Attachment Deleted
|
|
|
Post by tachyon on Apr 24, 2022 8:59:02 GMT
Amazing, it must be really interesting. How much of the improvement in football in the last 20 years do you put down to data? Also, how can people enter the world of football data analysis? Depends how you define improvement. Changed how it's played, definitely. Since data analytics became a thing in around 2010 open play shots have moved steadily closer to the goal. Around 2 yards closer, on average since 2014. That's all down to xG which measures the quality of goal scoring chances. In 2013 wrote how taking lots of low quality chances (from distance) is generally an inferior strategy to taking fewer hgher quality ones, even if in the aggregate both strategies yield the same total xG. It's all down to the higher variance of low quality opportunities. Football is a low scoring sport and sacrificing the slim chance of occasionally scoring 6 for a higher chance of more frequently scoring once is the way to go. Teams have told me that's influenced their shooting strategy. Fewer 30 yard screamers, but more points. Data has literally changed the game people watch. Crossing has also been transformed since pre smart Liverpool's disasterious foray into their Andy Carroll (4 goals) & Stewart Dowing (0 goals, 0 assists) led idiocy. Team's now cross from smart areas to smart areas (Tymon does it well for us, mostly). Downing didn't cross from the optimum position & keepers spent all day just catching them. Clubs are much more aware of ageing curves. How to plan for replacing the squad, how young players evolve in different positions and what's the benchmark for appearances/output spread over different leagues at different ages. I got into football data by just doing it, writing blogs about it (1 million current views, but it took me 8 years & 500 posts) & posting about it on Twitter. There's lots of free data out there, but in it's raw state it's just contextless trivia (see the current use of pass success to rate centre backs, which is wrong on so many levels). You need to turn it into actionable insight and that needs some statistical/coding/data science/ visualisation type skills. Shawcross' pass success rate was around 60% under Pulis & 80+ under Hughes. His passing ability didn't change over night, but his role did. He went from long balls to short passes, so you had to put his passing success rate into the context of the difficulty of each pass attempt he made under each manager.
|
|
|
Post by tachyon on Apr 23, 2022 11:53:50 GMT
Bristol City paid £1.8 million for Weimann Swansea paid £1.5 million for Piroe They have 39 league goals between them this season. Deals are certainly out there. Finding them is half scouting foresight and half blind luck I feel. Always exceptions and you need luck Generally though £5m does not buy you a 15 goal Championship striker That's literally what both teams have had. Weimann 19 non pen goals from 12.6 xG Piroe 19 non pen goals from 10.8 xG We know what's coming.
|
|
|
Post by tachyon on Apr 23, 2022 11:34:18 GMT
Based on all that, is the sensible thing to sell him this summer and use the money to buy a striker whose stats show he has underperformed in front of goal this season? In the same way we might have got a nice fee for Clucas after his golden season that he was never likely to repeat. Is that how the clubs like Leicester, Southampton and Brentford have been overachieving in the transfer market? Also, how can anyone spot the difference between a Jacob Brown at Barnsley, who gets into good positions and misses through variance, and a Dave Bamber type who gets into good positions and misses because he’s just a donkey? Re Clucas. I said on here he was never going to repeat his 2019/20 season. 11 goals from 5.8 xG and 3 from 4.1 xG subsequently. Same pattern, but in reverse. xG/90 stays roughly the same and massive over performance regresses to his xG after his purple patch. Goal scoring cannot escape the pull of reverting to your underlying process. Re Brentford's model. Find under valued players using data from geeks of Twitter, get good use out of some signings, then sell on to slightly gullible teams who play an entirely different system and haven't noticed their purchase also takes penalties. Disband Twitter geeks, but rinse/repeat in house until the stars align and you get promotion. Not sure what our long term plan is, so I'm unsure re JB. I'd lean to keep. Bamber has very little statistical data, just part myth from way back. But he did score around a goal every 3 games in a variety of divisions. Probably wasn't a Championship quality player, looked awkward, didn't get as many scoring opportunities up in class and the "he must score" football dinosaurs of the day tucked in. He scored a goal every 2 games in the (old) 4th division. A goal every 3.5 in the 3rd and a goal every 4.4 in the 2nd. A career stretching over 500 games suggests he was an ok finishers who hit a ceiling in the second tier.
|
|
|
Post by tachyon on Apr 23, 2022 9:34:30 GMT
So Brentford, the club that has been lauded more than any other for it's intelligent and consistently good recruitment, are stupid? And presumably you know better Embarrassed for you, your utter refusal to accept that your opinion of Brown is not only wrong but utterly risible Here's how Stoke *may* have picked up an absolute bargain when they signed JB. In his Championship season at Barnsley he scored three goals in 3568 minutes. That's a goal every 12.5 full games. But he got on the end of 8.3 xG. That's 1 expected goal every 4.5 full games. Here's his shot map. The bigger circles are excellent quality chances. He scored one from nine. But smart recruitment will know that getting into those good scoring positions is a very valuable skill and in cases of such under performance there's upwards of a 70% likelihood that your total actual goals scored will tend towards your xG in the future, rather than staying low. It's what mostly happens because variance is a huge cause of those missed chances, not as some think "being useless at football". So in deal terms you try to "buy" at the price that reflects JB's paltry three goals, but there's a really good chance that when you get him home you've actually bought a player that will score a goal every 4.5 games. A bargain & a very good player. And that's what has happened. We've got a player who has scored an actual goal every 3.5 full games. He's actually over performed his xG. He's been "clinical" in outdated football terminology or benefited from positive variance in modern parlance. In short, his xG per 90 at Stoke is almost exactly the same has his xG/90 was at Barnsley. It's just his statistically noisy conversion rate that has fluctuated. *If* we sell now, we play dumb and sell on his actual (and currently slightly hot) goals per game rate and we also throw in his incredibly good expected assists record at Barnsley, which has partly disappeared in his different role at Stoke.
|
|
|
Post by tachyon on Apr 23, 2022 6:48:46 GMT
This. Eyes + Brain = Cognitive bias. They are generally poor tools to judge the world. Virtually every criticism of using stats is a strawman argument, either through not fully understanding the metric or using statistical trivia, without context (pass completion rates etc), rather than valid models. A data lead approach is indispensable in football in 2022. At risk of derailing the thread, I'm fascinated to know how the data-led approach would work with a side as idiosyncratic as our Stoke team in the period 2007 - 2011. You have a striker in the Mama role whose main jobs are to press the opposition midfield and win headers. Goals don't even come into the top ten of his job requirements. The other striker is sort of expected to score goals, but his value really comes from his ability to read flick-ons, chase the ball into channels and win set pieces. He's basically a one-man attack. The wingers have to do the work of a full back. The full backs are centre backs. The central midfielders play more like a high defensive line than actual midfielders. The centre backs are expected to score more goals than the strikers The whole game plan revolves around not having the ball and letting the opposition take shots from a distance. The system appears to be aimless hoofing but has a surprisingly sophisticated set of routines that catch teams out. A throw-in is as valuable as any other set play, if only for the fact that it draws every opposition outfield player into the penalty area and yields set pieces. Knowing that Pulis was very fond of his data, one wonders if the sea change from Mama/Fuller to Adam/Crouch was led by a limited interpretation of certain numbers that looked good on paper but not so good in reality? I also wonder how you go about effectively scouting players who can fulfil the above roles? My biggest ponderance with regards to all of the data, though, is where this all leads to. Ultimately, won't we arrive at a world where all clubs have access to near enough the same data, same statistics and same processing methodologies? I'm guessing that even the top football clubs aren't going to fund a ten year computational intelligence research programme to do something clever that other clubs can't manage. At that point, every club is surely going to converge on the same limited pool of players and we probably end up going back to a world where judgement calls get made on the basis of individual biases. Pulis' Stoke was actually easier to have a succession plan for because of the fairly unique requirements. If you have an extensive database you can match (younger) similar players who you get in so that they can gradually oust the current encumbent who is then moved on while they still have some residual market value. That does assume a side has a recognisable and sustainable playing style and you'd also have a list of stylistically similar potential managerial replacements for continuity. That's not to say tweaks aren't needed (Delap replaced by two wingers). Liverpool absolutely have a long term research facility. Fully staffed, world leading data infrastructure, rolling player replacements, fully costed to ensure value for money, with a couple of potential Klopp replacements in mind should he depart. They've taken in £360 million in transfer fees in the last six seasons. They won't sell at under the odds & only buy at the right price. Shaq was only ever going to be a bit part player, but his release clause made him an attractive asset to have. (They recouped half that fee when he went to the MLS). Everything that is cutting edge in analytics, Liverpool can do. I sat next to their slightly reclusive Head of Research at a conference where Barca dscribed a novel, new technique to evaluate how players receive a pass. By the end of the talk he'd worked out what Liverpool needed to replicate it and a timeframe. (three days!).
|
|
|
Post by tachyon on Apr 22, 2022 16:42:58 GMT
Came across Lee in 1977 when we were both trying to impress George Eastham pre season. Nice kid, we warmed up by jogging around the Victoria Ground running track together. A team of scratch reserves/trialists beat the first team 1-0 in a 40 minute game on the practice pitches at the back of the Boothen End. Lee crossed it, I tapped it in at the far post. (Shilton, if I recall correctly). Just a flimsy excuse to boast about scoring the winner against Stoke's 77/78 team, really :-) I scored at Athletico Bilbao's ground as a kid but v no one famous 😄. A few weeks later Bryan Robson scored in the same goal v France. Mine was better. But was it as quick :-)
|
|
|
Post by tachyon on Apr 22, 2022 14:58:55 GMT
I thought Chapman was was fairly inept. He did score a few goals as he had some good players around him supplying good chances anyone could score, mainly two yards from goal Came across Lee in 1977 when we were both trying to impress George Eastham pre season. Nice kid, we warmed up by jogging around the Victoria Ground running track together. A team of scratch reserves/trialists beat the first team 1-0 in a 40 minute game on the practice pitches at the back of the Boothen End. Lee crossed it, I tapped it in at the far post. (Shilton, if I recall correctly). Just a flimsy excuse to boast about scoring the winner against Stoke's 77/78 team, really :-)
|
|
|
Post by tachyon on Apr 22, 2022 14:14:11 GMT
The problem is always when an opposing full back goes forward without getting tracked. DMs have to shift over, or one of them does, which then leaves an overload in the middle. The extra goal threat Campbell gives us when playing wide is, IMO, not worth it, as he leaves us so much more vulnerable at the back. (Awaits tachyon to post stats proving that opinion to be bollocks) 😂 You're on pretty safe ground when the only players below you in attempted tackles per 90 play in goal :-)
|
|
|
Post by tachyon on Apr 22, 2022 13:58:45 GMT
That could be because your eyes (as in everyone's) only see what they want to see. This. Eyes + Brain = Cognitive bias. They are generally poor tools to judge the world. Virtually every criticism of using stats is a strawman argument, either through not fully understanding the metric or using statistical trivia, without context (pass completion rates etc), rather than valid models. A data lead approach is indispensable in football in 2022.
|
|
|
Post by tachyon on Apr 21, 2022 17:50:46 GMT
So who models the data? Is it done in-house at every club? (which if so must provide a whole new level of inter-club warfare) Or do companies sell the modelling to multiple clubs? Or the highest bidder etc? Mostly in house. At Liverpool it's done by a nuclear physicist x2, a former chess champion of Wales, a particle physicist from CERN and a video game software developer, for Statsbomb it's a climate change scientist, for Man City it's a Harvard astro-physicist and former policy advisor to the Treasury, Coventry had an economist, as did Swansea, Arsenal, a data scientist, there's a smattering of professors of applied maths, some have landed jobs via Twitter (Luton) and Infogol/Sporting Life's player/team modelling is done by me :-). It's dog eat dog!
|
|